Design and Evaluate Procedures to Maintain Behavior Change
If you’ve just watched a learner nail a new skill in your clinic—perfect accuracy, minimal prompts, consistent execution—you might feel the urge to celebrate and move on. But here’s the real question: Will that skill stick around when your client goes home? When a new teacher takes over? When you’re not there to reinforce it?
This is where behavior maintenance procedures come in. Maintenance is about keeping a learned behavior at the desired level over time and across the situations where it matters most. It’s not a phase you add at the end of teaching. It’s something you plan into your intervention from day one, and it’s critical to your clients’ long-term success.
In this post, you’ll learn how to design and evaluate maintenance procedures that actually work in the real world—not just in your clinic. We’ll cover what maintenance really means, how to plan for it, common pitfalls to avoid, and how to make decisions based on data when things don’t go as expected.
What Is Behavior Maintenance? A Clear Definition
Maintenance (also called response maintenance) means a learned behavior continues after you’ve removed or significantly reduced the prompts, reinforcers, and special teaching conditions you used to teach it. It’s the ultimate goal: a skill that’s truly in a person’s repertoire, stable over time, and useful in everyday life.
Think of it this way. You teach a teenager to prepare a simple meal using step-by-step prompts and rewards. That’s acquisition. Maintenance is when that same teenager makes that meal reliably weeks later without your reminders—and without needing a prize every time. The behavior has become durable, independent, and real.
Maintenance has a few hallmarks. First, the learner can perform the skill consistently across days, weeks, or months. Second, they’re moving toward independence, relying less on prompts or special supports. Third, the behavior holds up even when artificial rewards stop. That’s resilience, and that’s what we’re after.
Maintenance Versus Other Key Concepts
It helps to know exactly how maintenance differs from other things you might be working on.
Maintenance versus generalization. These two concepts work together but are separate. Maintenance is about time: Does the skill stay strong over days and weeks? Generalization is about transfer: Can the learner use the skill with different people, in different places, or with different materials? Both matter. A behavior can generalize beautifully across three teachers but still fade away in two months if you don’t plan for maintenance. Smart programs plan for both from the start.
Maintenance versus acquisition. Acquisition is teaching the skill in the first place. Maintenance is keeping it. Acquisition uses intensive prompts and frequent reinforcement. Maintenance gradually removes those supports while checking that performance doesn’t drop.
Maintenance versus ongoing reinforcement. Some behaviors do need continuous support—a medication reminder, a safety routine requiring constant vigilance, or a learner who needs perpetual caregiver involvement. But most skills should eventually run on “fumes”—natural reinforcers, habit, or intrinsic motivation—rather than therapy-level reward schedules. Maintenance planning helps you figure out which is which.
Maintenance versus relapse. You hope for maintenance and plan for it. But sometimes skills do come back. Relapse is when a behavior that was maintained suddenly drops or disappears. That’s not failure; it’s a signal that your plan needs updating.
Why Maintenance Planning Matters—And What Goes Wrong Without It
Picture this: You’ve spent six months teaching a child to use an alternative communication device. They’re at 95% accuracy in your clinic, zero prompts needed, loving the system. You discharge them to their school. Three months later, the parents say the device is “just sitting in a drawer.” The skill hasn’t maintained. Time, money, and effort—gone.
This happens more often than it should, and it’s usually preventable.
Without a solid maintenance plan, skills regress. Natural supports aren’t in place. Caregivers weren’t trained. Reinforcement wasn’t properly thinned. The behavior depended on your presence, your prompts, or your rewards, and once those were gone, so was the skill.
But maintenance planning also serves a deeper purpose. A skill that actually lasts improves a person’s real life. Independence, community participation, self-care, communication—these things only matter if they stick around. Maintenance is about dignity and function, not just clinical progress.
From an ethical standpoint, you have a duty to plan for this. Your clients deserve interventions with staying power. That means planning maintenance early, involving natural supports, and using data to guide your decisions.
Core Components of a Maintenance Plan
A solid maintenance plan has several moving pieces.
Explicit maintenance goals. Don’t just say “maintain the skill.” Be specific: “Use the communication card with 90% accuracy across two different settings for at least five consecutive weeks.” Your goal should specify the performance level, the settings or people involved, and the timeframe.
Follow-up probe schedule. You can’t know if a skill is maintaining if you’re not checking. A typical schedule might be weekly probes for the first month after discharge, then biweekly for two months, then monthly for six months. Adjust based on risk—a safety skill might warrant longer or more frequent monitoring.
Fading and thinning strategy. This is your plan for gradually removing supports. Maybe you fade prompts using a most-to-least hierarchy. Maybe you thin reinforcement from every correct response down to every third response down to natural reinforcers only. The specifics depend on the behavior, but the principle is the same: gradual, deliberate, with data checks at each step.
Stakeholder training. Maintenance doesn’t happen in a vacuum. Parents, teachers, group home staff, or other caregivers need to understand the behavior, know how to respond, and have the tools to support it. A training plan specifies what you’ll teach them, how you’ll teach them, and who checks that they can do it.
Data collection plan. Specify what you’ll measure (accuracy? frequency? independence level?), how often, and where results go. This removes guesswork and keeps everyone accountable.
Decision rules. If a probe shows the skill is at 85% accuracy and your criterion is 90%, what do you do? If performance is great for two probes but drops on the third, is that a pattern or a fluke? Decision rules answer these questions in advance. For example: “If performance drops below criterion on two consecutive probes, reintroduce the targeted prompt for two weeks and re-probe.”
When to Start Planning for Maintenance
Many clinicians make a timing mistake: they think about maintenance only after the learner hits mastery. But that’s backward. Maintenance should be built into your plan from session one.
When you’re writing your initial behavior plan, include a section on how you’ll fade supports and what maintenance will look like. When the learner reaches criterion-level performance, actively shift into maintenance mode—start fading prompts, start thinning reinforcement, start training caregivers. Before discharge or a transition, finalize your probe schedule and decision rules. And during major life transitions—a move, a change in caregivers, a change in settings—revisit your maintenance plan and adjust if needed.
Practical Example: Transferring Reinforcement to Caregivers
Let’s make this concrete. A five-year-old has learned to request preferred items by pointing to pictures. In therapy, she got a high-value edible reward every time. Now you’re moving toward maintenance.
Your maintenance goal: “Initiate requests using picture board with 90% accuracy across home, school, and clinic for four consecutive weeks without tangible reinforcement.”
Your plan includes a gradual shift. For two weeks, the parent provides edibles on a continuous schedule. For the next two weeks, on a variable ratio-2. Then, for two weeks, you introduce natural reinforcers—the item she requested is the reward. Meanwhile, you’re probing weekly to make sure accuracy stays at 90% or above.
Your decision rule: “If any probe falls below 85%, the parent returns to the previous reinforcement density for one week, then attempts the fade again.”
After four weeks at 90% accuracy without tangible reinforcement, you move to biweekly probes for one month, then monthly for two months. If all probes hit criterion, you formally discharge.
Another Example: Fading Prompts and Coaching Staff
An adolescent has learned a three-step self-care routine (shower, dry off, get dressed) using physical and verbal prompts in a day program. Mastery was reached with verbal prompts only and 95% accuracy over five sessions.
Your maintenance goal: “Complete the three-step routine independently with 100% accuracy across five consecutive days.”
Your plan includes a prompt fade. You reduce from verbal prompts to gestures over one week. Then from gestures to no prompts, while a staff person watches from a distance. Meanwhile, you coach the staff on how to respond if the learner struggles: use the least intrusive prompt that works, and only if the person hasn’t started the next step after 10 seconds.
Probes happen daily during the fade, then drop to three times weekly for two weeks, then weekly for four weeks. Your decision rule: “If the learner needs more than one verbal prompt on three probes, pause the fade, return to gestures, and try again in one week.”
You also hold a brief staff training session every two weeks. After criterion is met, you schedule formal probes at one month, three months, and six months post-discharge. The staff knows that if the behavior dips below 90% accuracy on any probe, they should contact you for troubleshooting.
Common Mistakes and How to Avoid Them
Here’s where many clinicians go off the rails.
Planning maintenance only after mastery. By then, you’ve designed your whole treatment without thinking about fade-out. Solution: Include a “maintenance section” in every behavior plan from the start.
Fading supports too fast. It’s tempting to pull back all prompts or rewards at once to “see if they’ve really got it.” They almost never do. Solution: Use a written fade plan with specific timeframes. Let data guide adjustments, not impatience.
Stopping data collection after discharge. That’s exactly when you should start your maintenance probes. Solution: Pre-schedule your entire probe timeline before discharge. Make it non-negotiable.
Designing maintenance that only works with you. You fade prompts but the reinforcement is still your praise, your reward, your schedule. When the learner leaves, the skill collapses. Solution: From the start, use natural reinforcers and involve caregivers in prompting and reinforcement.
Confusing maintenance probes with generalization probes. A maintenance probe checks if the skill holds up over time in routine conditions. A generalization probe checks if the skill works with a new teacher or in a new location. Solution: Know what you’re testing. If you’re checking maintenance, use familiar people and settings.
Using Data to Make Maintenance Decisions
Data is your guide when things don’t go as planned.
Suppose you’re three weeks into maintenance probes and the last two probes show 78% and 75% accuracy. Your criterion is 90%. Before you panic, ask yourself: Was something different about those probes? Did the caregiver implement differently? Did the environment change? One low probe might be noise. Two in a row suggest a real issue.
Your decision rule should tell you what to do. Let’s say it said: “If performance drops below 85% on two consecutive probes, reintroduce the most recent faded support for one week and resume probing.” Now you reintroduce a bit of verbal prompting, coach the caregiver again, and probe daily for a week. If performance bounces back to 90%+, you resume the fade plan but more slowly. If it stays low, you reassess: Is the criterion too high? Is the caregiver able to implement with fidelity? Does the learner need more time?
This is not failure. This is how maintenance planning works. You gather data, follow your rules, adjust, and keep the learner’s best interest at the center.
Ethical Foundations of Maintenance Planning
Maintenance procedures involve fading reinforcement and supports. That means changes in how the learner experiences their environment. This requires ethical care.
Start with informed consent. Before you fade reinforcement or prompts, talk with the learner (if they can understand) and their caregivers. Explain what will change, why, and what to expect. Mention that sometimes behavior can increase temporarily when rewards are faded—an extinction burst—and that this is normal and manageable.
Use reinforcement thinning deliberately and gradually. Jumping from continuous reinforcement to very lean reinforcement can cause ratio strain: the learner gives up because the reward feels too far away. Thin gradually. Monitor for signs of frustration or decreased responding.
Prioritize least-restrictive supports. If natural reinforcers can maintain a skill, don’t keep using therapist-delivered rewards. If a gesture can replace a verbal prompt, do it. Independence and dignity are the goals.
Ensure equity and practicality. Some caregivers have limited time, knowledge, or resources. A maintenance plan that expects a busy parent to probe twice a week will fail. Be realistic about what caregivers can do and provide the training and tools they need.
Document everything. Write down your maintenance goal, your probe schedule, your decision rules, and your reasoning. If something changes later, you’ll have a clear record of what was planned and why.
Monitoring for Relapse and Relapse Prevention
Even with a solid plan, sometimes behaviors regress. A learner might lose a skill months or years after successful maintenance. This doesn’t mean you failed. It means the situation changed in a way your plan didn’t anticipate.
Common triggers for relapse include a change in caregivers, a move to a new environment, stress, illness, or a change in reinforcement availability. A good maintenance plan anticipates these. You might note: “If the learner’s primary caregiver changes, schedule a booster session and resume weekly probes for one month.”
If relapse does happen, treat it as a data point, not a disaster. Review your records. What changed? Reintroduce supports at a level that works, gather data, and adjust. This is problem-solving, and it’s an essential part of real-world clinical work.
Documenting Maintenance in Your Behavior Plan
When you write a behavior plan, include a clear maintenance section.
Start with your maintenance goal: a sentence or two defining the target performance, accuracy level, settings involved, and timeframe. “Client will initiate requests using the AAC device with 90% accuracy, with three different communication partners, in home and school settings, for four consecutive weeks” is clear. “Client will maintain the skill” is not.
Next, your probe schedule. A simple table works well: probe frequency, duration, and the exact measures you’ll use. Be specific about who will conduct probes and where.
Then, your fading and reinforcement-thinning plan. Write the steps, timeline, and how you’ll know when to move to the next step. Include your decision rule: “If performance remains at 90%+ for two consecutive probes, thin reinforcement to variable ratio-2.”
Add a caregiver training plan. What will you teach? When? How will you check that they’ve learned?
Finally, list your decision rules clearly. These let everyone—you, the team, the caregivers—know exactly what happens next.
Key Takeaways
Maintenance is keeping learned behaviors at the desired level over time under routine conditions. Build it into your plan from the start, not as an afterthought. Define clear maintenance goals, include a realistic probe schedule, and involve natural supports and caregivers as active partners.
Use data to guide your fading and thinning decisions. Don’t move too fast, and don’t stop collecting data once the client leaves your clinic. Set clear decision rules so everyone knows what to do if something changes.
Remember that maintenance is about real-world function and dignity. A skill isn’t truly learned if it disappears when your supports are gone. By planning and monitoring carefully, you ensure your clients’ gains are durable and their independence grows.
